We are happy to announce CDS 2.3 release 3 Powered by Apache Spark. You can download the parcel and apply it directly to provisioned clusters without disrupting your currently running Spark workloads.
This component is generally available and is supported on CDH 5.9 and higher, and Cloudera Manager 5.11 and higher.
This is purely a maintenance release and it includes all fixes that are in the Apache Spark 2.3.1 upstream release. Test-only changes are omitted. For more information, see theApache Spark 2.3.1 upstream release notes.
SPARK-16451- [REPL] Spark-shell / pyspark should finish gracefully when "SaslException: GSS initiate failed" is hit
SPARK-17756- [PYTHON][STREAMING] java.lang.ClassCastException returned when using 'cartesian' with DStream.transform
SPARK-24029- Set the "reuse address" flag on listen sockets
SPARK-24216- [SQL] Spark TypedAggregateExpression uses getSimpleName this is not safe in Scala
SPARK-24369 - [SQL] Correct handling for multiple distinct aggregations that have the same argument set
SPARK-24468- [SQL] DecimalType 'adjustPrecisionScale' might fail when scale is negative
SPARK-24495- [SQL] SortMergeJoin with duplicate keys produces wrong results
SPARK-24506- [UI] Add UI filters to tabs added after binding
SPARK-24542- [SQL] Hive UDF series UDFXPathXXXX allows users to pass carefully crafted XML to access arbitrary files
SPARK-24548- [SQL] JavaPairRDD to Dataset<Row> in Spark generates ambiguous results
SPARK-24552 - Task attempt numbers are resused when stages are retried